Wind tunnel or sound funnel? (Image: Shaun Jackson/)

For six weeks this summer, coinciding with the , gigantic, unprepossessing buildings with names like Q121 and R52 are humming, droning and singing in celebration. It’s all part of that utilises the buildings’ exceptional acoustic properties.

The buildings in Hampshire, UK, are the wind tunnels that shaped the Spitfire’s peculiar, elliptical wings, as well as guiding designs for supersonic aircraft from as far back as 1942. Now they are for the first time.

Finished in 1935, Q121 is a steel and reinforced concrete building some 15 metres high, housing Britain’s largest wind tunnel. Designed to channel air in the most efficient manner, the tunnel boasts extraordinary acoustics. The machine’s massive fan once drove air at 185 kilometres an hour into the maw of a concrete throat. Reinforced concrete vanes turned the gale back on itself twice over until the airstream left through a grating aimed directly at the intake fan. The open test area between the vent and the fan was large enough to accommodate entire planes.

The reverberation is tremendous, of course: make a sound in any large industrial building and it lingers in the space. In acoustic engineering, the rate of decay is gauged by the reverberation time, which is the time it takes for sound to die away to 1/1000th of its initial intensity. The reverberation time is largely controlled by the amount of energy dissipated every time sound bounces off the walls, floor and ceiling. Larger spaces have longer reverberation times because there is a longer interval between reflections.

In Q121, though, something else is happening as well. “The structure is built precisely to minimise turbulence, so along with the reverb you get enormous clarity,” says McIntyre-Burnie. His installations here combine instrumental music, historical recordings and the voices of former engineers. “Also, the spaces between the vanes that directed the air in the tunnel seem to have their own acoustic properties.”

Whispering walls

You might think that sound art would be easy to record, but it presents some unique problems. “Being in a space, gauging the distance between sound sources, walking through corridors and rooms, you subconsciously compensate for the way sounds from different locations arrive at different times in the ear,” says McIntyre-Burnie. “Installations can make perfect sense in situ, but they can come apart in a recording, when there are no visual cues to help you.”

His current installation is a case in point. To one side of the test area, where the main speakers hang, there is an access door leading to another part of the tunnel. Sounds travelling by this route arrive after an appreciable delay.

This aspect of sound art is, as yet, not very well understood, and since no one is going to build cathedral-like spaces for artists like Thor McIntyre to play in, that’s all the more reason to open up derelict and disused industrial spaces to artistic exploration. , sites with extraordinary acoustics deserve preservation just as much as those with remarkable visuals.

For another example of such special acoustics, look to Teufelsberg, on the outskirts of Berlin, where there is a disused military facility containing a number of “radomes” – spheres used to protect and conceal radar antennas. The highest radome is on the sixth storey of a derelict tower. Jump onto the concrete plinth in the centre of the room, and any sound you make is focused straight back towards you. Try swaying to the right so the structure’s focal point is at your left ear; now you can whisper into your own ear, using the amplification afforded by the curved walls.

Could these experiences be preserved by marrying 3D audio recordings from such spaces to visual virtual reality systems? Perhaps, but it remains to be seen whether such recordings would ever rival the magic of actually being there.

The Wind Tunnel Project runs until 20 July 2014, from 10 am to 8 pm.

Last week, I invited readers of żěè¶ĚĘÓƵ to take part in to see whether acousticians had succeeded in programming a computer to fake a human laugh.

Over 6000 of you did so. The result? That it takes a lot to fool you – although the computers might be getting the upper hand.

As a series of articles in żěè¶ĚĘÓƵ last week showed, laughter is essential to expressing many facets of human behaviour and emotion – not just that you appreciate a joke. No surprise, then, that how to fake it convincingly has been on people’s minds for quite some time.

As a homage to the early pioneers of speech synthesis, one of the laughs we used was made by a replica of a developed by the inventor Wolfgang von Kempelen in 1791. Times and tastes change, and perhaps unsurprisingly 85 per cent of you spotted that as a fake.

Audio: Von Kempelen speaking machine

Virtual anatomy

What difference have 200 years of development in speech synthesis made? In some cases, it seems, rather little. and colleagues from the University of Saarland in Saarbrücken, Germany, for example, recreate laughter using computer models to simulate movements of the vocal tract and air flow. An impressive 88 per cent of you correctly identified this example as synthesised laughter.

Audio: Jürgen Trouvain’s synthesised laugh

An algorithm used by of the Pierre and Marie Curie University in Paris, France, does not fare any better. It turns text into speech and then alters the patterns of stress and intonation to try to convey different emotions. Over 90 per cent of you spotted these following two sounds as fakes – but as a supplementary question, can you tell which is meant to be happy and which sad? (Answer at bottom of article.)

Audio: Grégory Beller’s synthesised laugh (one)

Audio: Grégory Beller’s synthesised laugh (two)

The technique used by at Deutsche Telekom in Berlin, Germany, had a little more success. He uses linear predictive speech coding to generate individual laughs (“ha”), and a simple algorithm to string a series of these laughs together. We threw seven examples into the mix. On average, 78 per cent of you spotted that sounds such as the following were synthesised.

Audio: Shiva Sundaram’s synthesised laugh

Part human, part machine

So you might be feeling a little smug so far – but not so fast. Our experiment did provide a clear winner: one laugh that had almost 60 per cent of you thinking it could be human. It was produced by from the University of Mons, Belgium, using a hybrid method of sampling and synthesis, mixing and manipulating single laughs drawn from real laughter. So here is the computer-generated laugh that had many of you fooled.

Audio: Jérôme Urbain’s synthesised laugh

Arguably, complete success will only be achieved when the vast majority of people think a synthesised laugh is human – or, perhaps more realistically, about 80 per cent of people, given that only 80 per cent of respondents to the test identified human laughs as human. Perhaps that is not so far away.

And what of the other “secrets of laughter” covered in żěè¶ĚĘÓƵ? Does our experiment provide any evidence, for example, for the idea that men and women interpret laughter differently? Not really: although the men have it by a whisker, identifying laughs correctly 76 per cent of the time as opposed to 72 per cent of the time for women, the differences are small.

And because we weren’t recording you while you did the test, we can’t report whether any of you were seized by the uncontrollable desire to laugh yourselves.

Happy or sad?

The top sound was sad and the bottom one happy.

Love ’em or loathe ’em, the blaring plastic trumpets have become the hallmark of matches at the 2010 World Cup. We asked , president of the UK Institute of Acoustics and an acoustic engineer at the University of Salford, UK, to explain their appeal – or otherwise

How do vuvuzelas make their sound?

The vuvuzela is like a straightened trumpet and is played by blowing a raspberry into the mouthpiece. The player’s lips open and close about 235 times a second, sending puffs of air down the tube, which excite resonance of the air in the conical bore. A single vuvuzela played by a decent trumpeter is reminiscent of a hunting horn – but the sound is less pleasing when played by the average football fan, as the note is imperfect and fluctuates in frequency. It sounds more like an elephant trumpeting. This happens because the player does not keep the airflow and motion of the lips consistent.

But that din sounds nothing like a trumpet or an elephant.

When hundreds of the vuvuzelas are played together, you get the distinctive droning sound. People in the crowd are blowing the instrument at different times and with slightly varying frequencies. The sound waxes and wanes. The overall effect is rather like the sound of a swarm of insects.

Why are they so loud?

The loudness can be explained by the bore shape, which is roughly conical, and flares. As well as creating sound at a frequency of 235 hertz, the instrument generates harmonics – sound at multiples of the fundamental frequency. We have measured strong harmonics at 470, 700, 940, 1171, 1400 and 1630 hertz.

A flared instrument has louder higher-frequency harmonics than a cylindrical one. The flared instrument is perceived as louder because the higher harmonics are at frequencies where our hearing is most sensitive. This is partly why the conical saxophone sounds louder than the cylindrical clarinet.

Since it produces 116 decibels at 1 metre, , according to a study by the Department of Communication Pathology at the University of Pretoria, South Africa. Listen to just one instrument for 7 to 22 seconds and you exceed typical permitted levels for noise at work. A whole crowd produces even higher levels, and measurements at a training match have shown temporary hearing loss among spectators.

Is it annoying because it is loud?

Experiments on other noise sources show that louder sounds are more annoying. Our hearing is an early-warning system: we listen out for sudden changes in the sounds around us which might indicate threats, and ignore benign, persistent noise. When noise becomes as loud as a vuvuzela, however, it becomes impossible to habituate to the sound.

What else about the sound makes it annoying?

The droning quality makes it more annoying – the fact it has a distinct pitch or note. Investigations into many noise annoyance problems have demonstrated this. Indeed some noise standards and regulations have corrections to allow for the additional annoyance from such sound. Droning sounds are harder to ignore and more alerting than broadband noise such as the hiss of a badly tuned radio. This might be because tones can carry useful information in the vowel sounds of speech. But it might also relate to threat detection – because predator sounds like a lion’s roar has tonal components – but I’m speculating.

What can be done to make it less annoying, especially on TV and the radio?

Broadcasters have to balance how much crowd sound to use compared to the commentators’ voices. If they make the crowd too quiet then the game lacks atmosphere, so they can’t turn it off altogether. If you are watching the match on a computer, at Queen Mary University of London. Otherwise, you might just have to try and accept the sound as being part of the background. Lack of control over a noise source has been shown to increase its perceived annoyance. So your best bet might be to crack open another beer and try your best to enjoy the atmosphere.